Focused Library Generator: case of Mdmx inhibitors

We present a Focused Library Generator that is able to create from scratch new molecules with desired properties. After training the Generator on the ChEMBL database, transfer learning was used to switch the generator to producing new Mdmx inhibitors that are a promising class of anticancer drugs. Lilly medicinal chemistry filters, molecular docking, and a QSAR IC 50 model were used to refine the output of the Generator. Pharmacophore screening and molecular dynamics (MD) simulations were then used to further select putative ligands. Finally, we identified five promising hits with equivalent or even better predicted binding free energies and IC 50 values than known Mdmx inhibitors. The source code of the project is available on https://github.com/bigchem/online-chem .

[1]  J. Reymond.  The chemical space project. , 2015, Accounts of chemical research.

[2]  Sergey Nikolenko,et al.  druGAN: An Advanced Generative Adversarial Autoencoder Model for de Novo Generation of New Molecules with Desired Molecular Properties in Silico. , 2017, Molecular pharmaceutics.

[3]  Koji Tsuda,et al.  ChemTS: an efficient python library for de novo molecular generation , 2017, Science and technology of advanced materials.

[4]  Tingjun Hou,et al.  Assessing the Performance of the MM/PBSA and MM/GBSA Methods. 1. The Accuracy of Binding Free Energy Calculations Based on Molecular Dynamics Simulations , 2011, J. Chem. Inf. Model..

[5]  Constantine Bekas,et al.  “Found in Translation”: predicting outcomes of complex organic chemistry reactions using neural sequence-to-sequence models† †Electronic supplementary information (ESI) available: Time-split test set and example predictions, together with attention weights, confidence and token probabilities. See DO , 2017, Chemical science.

[6]  John P. Overington,et al.  ChEMBL: a large-scale bioactivity database for drug discovery , 2011, Nucleic Acids Res..

[7]  Yoram Singer,et al.  Adaptive Subgradient Methods for Online Learning and Stochastic Optimization , 2011, J. Mach. Learn. Res..

[8]  G. Bemis,et al.  The properties of known drugs. 1. Molecular frameworks. , 1996, Journal of medicinal chemistry.

[9]  Igor V. Tetko,et al.  Online chemical modeling environment (OCHEM): web platform for data storage, model development and publishing of chemical information , 2011, J. Cheminformatics.

[10]  D. Lane,et al.  What the papers say: The p53‐mdm2 autoregulatory feedback loop: A paradigm for the regulation of growth control by p53? , 1993 .

[11]  Yan Zhao,et al.  Diaryl- and triaryl-pyrrole derivatives: inhibitors of the MDM2–p53 and MDMX–p53 protein–protein interactions† †Electronic supplementary information (ESI) available: Experimental details for compound synthesis, analytical data for all compounds and intermediates. Details for the biological evaluatio , 2013, MedChemComm.

[12]  T. Holak,et al.  Structure of the human Mdmx protein bound to the p53 tumor suppressor transactivation domain , 2008, Cell cycle.

[13]  Anang A. Shelat,et al.  First Small Molecule Inhibitor of MDMX Identification and Characterization of the Cell Biology : , 2010 .

[14]  Thomas Blaschke,et al.  Molecular de-novo design through deep reinforcement learning , 2017, Journal of Cheminformatics.

[15]  Igor V. Tetko,et al.  Chemical space exploration guided by deep neural networks , 2019, RSC advances.

[16]  Eric J. Martin,et al.  In silico generation of novel, drug-like chemical matter using the LSTM neural network , 2017, ArXiv.

[17]  Jürgen Schmidhuber,et al.  Long Short-Term Memory , 1997, Neural Computation.

[18]  Yoshiyuki Hirata,et al.  Discovery of new low-molecular-weight p53-Mdmx disruptors and their anti-cancer activities. , 2016, Bioorganic & medicinal chemistry.

[19]  Jonas Boström,et al.  Deep Reinforcement Learning for Multiparameter Optimization in de novo Drug Design , 2019, J. Chem. Inf. Model..

[20]  Joseph Schoepfer,et al.  Structural States of Hdm2 and HdmX: X‐ray Elucidation of Adaptations and Binding Interactions for Different Chemical Compound Classes , 2019, ChemMedChem.

[21]  Jean-Louis Reymond,et al.  Drug Analogs from Fragment-Based Long Short-Term Memory Generative Neural Networks , 2018, J. Chem. Inf. Model..

[22]  Guillermina Lozano,et al.  Molecular Pathways: Targeting Mdm2 and Mdm4 in Cancer Therapy , 2012, Clinical Cancer Research.

[23]  Matt J. Kusner,et al.  Grammar Variational Autoencoder , 2017, ICML.

[24]  Y. Haupt,et al.  The long and the short of it: the MDM4 tail so far , 2019, Journal of molecular cell biology.

[25]  Igor V. Tetko,et al.  Neural network studies, 1. Comparison of overfitting and overtraining , 1995, J. Chem. Inf. Comput. Sci..

[26]  Wei Gu,et al.  Is MDMX the better target? , 2018, Aging.

[27]  Arthur J. Olson,et al.  AutoDock Vina: Improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading , 2009, J. Comput. Chem..

[28]  Holger Gohlke,et al.  The Amber biomolecular simulation programs , 2005, J. Comput. Chem..

[29]  Kunio Arai,et al.  Automatic generation of synthetic routes from monosaccharides , 2007 .

[30]  Michael K. Gilson,et al.  BindingDB in 2015: A public database for medicinal chemistry, computational chemistry and systems pharmacology , 2015, Nucleic Acids Res..

[31]  I. Tetko,et al.  Applicability domain for in silico models to achieve accuracy of experimental measurements , 2010 .

[32]  John K. Buolamwini,et al.  Therapeutic Discovery A Small-Molecule Inhibitor of MDMX Activates p 53 and Induces Apoptosis , 2011 .

[33]  Michael M. Madden,et al.  Synthesis of cell-permeable stapled peptide dual inhibitors of the p53-Mdm2/Mdmx interactions via photoinduced cycloaddition. , 2011, Bioorganic & medicinal chemistry letters.

[34]  Toshiaki Hara,et al.  N-acylpolyamine inhibitors of HDM2 and HDMX binding to p53. , 2009, Bioorganic & medicinal chemistry.

[35]  Igor V. Tetko,et al.  GEN: highly efficient SMILES explorer using autodidactic generative examination networks , 2019, Journal of Cheminformatics.

[36]  Petra Schneider,et al.  Generative Recurrent Networks for De Novo Drug Design , 2017, Molecular informatics.

[37]  Andrea Carotti,et al.  Expanding the horizon of chemotherapeutic targets: From MDM2 to MDMX (MDM4) , 2011 .

[38]  Jun Xu,et al.  QBMG: quasi-biogenic molecule generator with deep recurrent neural network , 2019, Journal of Cheminformatics.

[39]  John Lowe,et al.  Faculty Opinions recommendation of Discovery of a Dihydroisoquinolinone Derivative (NVP-CGM097): A Highly Potent and Selective MDM2 Inhibitor Undergoing Phase 1 Clinical Trials in p53wt Tumors. , 2015 .

[40]  A. Levine,et al.  The mdm-2 oncogene product forms a complex with the p53 protein and inhibits p53-mediated transactivation , 1992, Cell.

[41]  Colin Raffel,et al.  Lasagne: First release. , 2015 .

[42]  A. Levine,et al.  Surfing the p53 network , 2000, Nature.

[43]  David E. Shaw,et al.  PHASE: a new engine for pharmacophore perception, 3D QSAR model development, and 3D database screening: 1. Methodology and preliminary results , 2006, J. Comput. Aided Mol. Des..

[44]  Ian A. Watson,et al.  Rules for identifying potentially reactive or promiscuous compounds. , 2012, Journal of medicinal chemistry.

[45]  John K. Buolamwini,et al.  A Small-Molecule Inhibitor of MDMX Activates p53 and Induces Apoptosis , 2010, Molecular Cancer Therapeutics.

[46]  Maria M. M. Santos,et al.  An Update on MDMX and Dual MDM2/X Inhibitors. , 2018, Current topics in medicinal chemistry.

[47]  E. Revilla,et al.  Individual Spatial Responses towards Roads: Implications for Mortality Risk , 2012, PloS one.

[48]  Alán Aspuru-Guzik,et al.  Automatic Chemical Design Using a Data-Driven Continuous Representation of Molecules , 2016, ACS central science.

[49]  Chandra S Verma,et al.  Differential binding of p53 and nutlin to MDM2 and MDMX: Computational studies , 2010, Cell cycle.

[50]  Alex Graves,et al.  Generating Sequences With Recurrent Neural Networks , 2013, ArXiv.

[51]  Jean-Louis Reymond,et al.  Enumeration of 166 Billion Organic Small Molecules in the Chemical Universe Database GDB-17 , 2012, J. Chem. Inf. Model..

[52]  Wei Wang,et al.  Structures of low molecular weight inhibitors bound to MDMX and MDM2 reveal new approaches for p53-MDMX/MDM2 antagonist drug discovery , 2010, Cell cycle.

[53]  Esben Jannik Bjerrum,et al.  SMILES Enumeration as Data Augmentation for Neural Network Modeling of Molecules , 2017, ArXiv.

[54]  Igor V. Tetko,et al.  Augmentation Is What You Need! , 2019, ICANN.

[55]  Ettore Novellino,et al.  Computer-Aided Identification and Lead Optimization of Dual Murine Double Minute 2 and 4 Binders: Structure-Activity Relationship Studies and Pharmacological Activity. , 2017, Journal of medicinal chemistry.

[56]  M. Oren,et al.  Mdm2 promotes the rapid degradation of p53 , 1997, Nature.

[57]  Igor V. Tetko,et al.  Transformer-CNN: Fast and Reliable tool for QSAR , 2019, ArXiv.

[58]  J. Levine,et al.  Surfing the p53 network , 2000, Nature.

[59]  Zhe Li,et al.  Moracin M from Morus alba L. is a natural phosphodiesterase-4 inhibitor. , 2012, Bioorganic & medicinal chemistry letters.

[60]  Tapan Behl,et al.  Reactivation of p53 gene by MDM2 inhibitors: A novel therapy for cancer treatment. , 2019, Biomedicine & pharmacotherapy = Biomedecine & pharmacotherapie.

[61]  A. Jochemsen,et al.  MDMX: a novel p53‐binding protein with some functional properties of MDM2. , 1996, The EMBO journal.

[62]  Olexandr Isayev,et al.  Deep reinforcement learning for de novo drug design , 2017, Science Advances.

[63]  Michael A. Dyer,et al.  On the Mechanism of Action of SJ-172550 in Inhibiting the Interaction of MDM4 and p53 , 2012, PloS one.

[64]  Joerg Kallen,et al.  Discovery of a Dihydroisoquinolinone Derivative (NVP-CGM097): A Highly Potent and Selective MDM2 Inhibitor Undergoing Phase 1 Clinical Trials in p53wt Tumors. , 2015, Journal of medicinal chemistry.

[65]  Holger Gohlke,et al.  MMPBSA.py: An Efficient Program for End-State Free Energy Calculations. , 2012, Journal of chemical theory and computation.

[66]  Xin Lu,et al.  Live or let die: the cell's response to p53 , 2002, Nature Reviews Cancer.

[67]  John Salvatier,et al.  Theano: A Python framework for fast computation of mathematical expressions , 2016, ArXiv.

[68]  Steven L Dixon,et al.  PHASE: A Novel Approach to Pharmacophore Modeling and 3D Database Searching , 2006, Chemical biology & drug design.

[69]  Eun Hye Kim,et al.  XI-011 enhances cisplatin-induced apoptosis by functional restoration of p53 in head and neck cancer , 2014, Apoptosis.

[70]  G. Wahl,et al.  Regulating the p53 pathway: in vitro hypotheses, in vivo veritas , 2006, Nature Reviews Cancer.

[71]  David Weininger,et al.  SMILES, a chemical language and information system. 1. Introduction to methodology and encoding rules , 1988, J. Chem. Inf. Comput. Sci..

[72]  Shinya Oishi,et al.  Affinity-based screening of MDM2/MDMX-p53 interaction inhibitors by chemical array: identification of novel peptidic inhibitors. , 2013, Bioorganic & medicinal chemistry letters.

[73]  Thierry Kogej,et al.  Generating Focused Molecule Libraries for Drug Discovery with Recurrent Neural Networks , 2017, ACS central science.